Do Gaps Always Fill? What Seven Years of Futures Market Data Reveals

Few claims appear in trading material as often as "gaps always fill." It is repeated in books, in chatrooms, and in social media threads, almost always without conditions and almost never with evidence. So we tested it — across Nasdaq-100 Futures, S&P 500 Futures, Gold Futures, Silver Futures, Crude Oil Futures, and Copper Futures, using approximately seven years of historical session data.

The short answer: gaps do fill, but not in the way most traders assume. And the behaviour, while real, did not translate into a reliable trading edge in any of the eight strategy implementations we tested.

The full paper is here: [Testing the "Gaps Always Fill" Myth Across Six Futures Markets](/research/gaps-always-fill-myth-futures-markets). This post is the plain-English version.

Why traders believe gaps always fill

Two reasons. First, small gaps really do fill quickly — in equity index futures especially, a tiny overnight gap is closed within minutes most of the time. Anyone who watches a chart at the open sees this happen over and over, and the brain generalises. Second, the claim is operationally specific in a way that traders like: open away from yesterday's close, take the trade back to yesterday's close, done. Specific rules feel like edge.

The problem is that the small-gap experience is being projected onto large gaps, where the behaviour is materially different.

Why gap size matters

A small gap usually reflects overnight noise: thin liquidity in adjacent sessions, ordinary order-flow imbalance, carry-over from the previous close. The prior settlement is still a credible anchor, and price drifts back to it.

A large gap usually reflects information: an earnings release, an inventory print, a geopolitical event, a central bank statement. The market isn't mispricing — it's repricing. There is no economic reason for price to return to a level the new information has invalidated, at least not in the same session.

That's why the headline statistic in the paper — same-session fill rate — drops sharply as gap size increases. Small gaps fill almost always. Medium gaps fill the majority of the time. Large gaps fill a minority of the time. Extreme gaps almost never fill the same day.

Why same-day statistics can be misleading

If you stop at "large gaps don't fill the same day" you've answered the wrong question. The question most traders actually care about is "will the gap fill eventually, and roughly how long does it take?"

When we extend the horizon, the picture changes. Large gaps display short-term persistence — they often continue to drift in the direction of the gap for a session or two before the reversion phase kicks in. But after that, fill probabilities increase steadily. By the end of the first trading week, a majority of large gaps in every market in our sample have filled. By approximately the tenth trading session, the large majority have filled.

That's the central empirical finding: gap-fill behaviour is real, but it's two-phase. Persistence first, then mean reversion.

Why persistence and mean reversion can both be true

There's no contradiction. Different participants act on different time horizons. Short-term flow follows the news that created the gap; longer-term participants who anchor to the prior settlement re-engage once the news has been digested. Both forces are visible in the data, and the dominant one depends on how many sessions you give the trade.

This also explains why the simplest version of the claim — fade every gap at the open, target the prior close — fights the persistence phase and loses on the trades where the persistence is strongest. The naive same-day fade is the wrong end of the rope.

Why a market tendency does not automatically create a profitable strategy

This is the part that surprises people most. We tested eight strategy implementations on the same dataset and the same six markets:

•same-session fades at the open

•delayed fades that waited one to three sessions before entering

•persistence-following trend variants for the first session

•multi-session swing fades with horizons up to ten sessions

•combinations filtered by gap size, by volatility, and by direction

Every variant was evaluated with realistic transaction costs and slippage. None of the eight produced a statistically reliable edge after costs. Some had high win rates but small wins relative to their losing tail. Some had positive raw expectancy that didn't survive cost assumptions. Some worked in one market and broke in another.

A pattern is not an edge. The gap-fill behaviour is real and measurable, but it isn't large enough or clean enough — on its own — to be a strategy.

The study findings, in plain English

1. Small gaps fill quickly, almost always within the same session.

2. Large gaps display short-term persistence and may continue in the direction of the gap for a few sessions.

3. Fill probability for large gaps increases steadily over time.

4. Most large gaps fill within approximately ten trading sessions across all six markets we tested.

5. The behaviour is statistically observable but did not translate into a reliable trading edge in any of the eight strategy implementations we tested.

In one sentence: gaps tend to fill, but the obvious way to trade that tendency does not work.

What this means for how we work at MPM

This is the kind of question MPM Markets runs before evaluating concepts or building models. Independent quantitative research first; a measurement to replace an assumption. The full paper, with the methodology, definitions, horizon tables, and the eight strategy variants, is below.

[Read the Full Research Paper →](/research/gaps-always-fill-myth-futures-markets)

If you want to see how this kind of testing feeds into the day-to-day framework, the [Backtest Library](/backtest-library) and [How It Works](/how-it-works) pages are the natural next reads.